Aerial Precision Drop

Back to Home



During the last two years of my stay at MIT, I was involved in a project, "Aerial Deployment of Sensor Networks from Stand-off Distances", which was the 3rd MIT/Draper Partnership Program (MDPP). As a part of the project, precision air drop vehicle was explored, which is supposed to deliver sensor and communication nodes from unmanned air vehicle (UAV) to ground. Two types of vehicle platform were considered for prototype development - ballistic drop, high-g platform and parafoil type low-g platform. I participated in the development and testing of the ballistic type vehicle.

Drop Vehicle

A couple of small sized ballistic drop vehicles were built. The overall length is about 60 cm and the total weight is 1.8 kg. The body has a cylindrical shape and it has four control fins. The four fins control the vehicle roll motion and the pitch and yaw attitude so that lift and side force are generated from the cylindrical body to move toward the target location as it free-falls. There is a parachute compartment at the end of the fin section. The parachute opens just before landing to reduce the speed of impact to ground during our testing. But opening the parachute itself turns out to create 30-40 g's. And because of the this reason, the important main structural part was made out of fiberglass tubing core, which runs all the way through the body and is attached to the parachute main string. A streamer is attached to the parachute and is exposed through a small hole made by the two doors of the parachute compartment. The door opens due to the pulling force created by the drag of the streamer.

Avionics

Inside the body section there are avionics subsystems. At the front there is a CCD camera which shows the view of the drop zone as the vehicle falls. This camera image is transmitted to ground station for a ground operator to identify and designate the target zone in real time. The main body is divided into two sections. On one side there is a main computer which is a PC104 CPU module with wireless LAN card. In the rear portion of the body on this side is a analog conditioning board with inertial sensors, which include three axis accelerometers and rate gyros. On the other side there is a battery package at the front followed by power supply unit and wireless video image transmitter. In the rear portion on this side there is a micro-computer that is dedicated to the control of servo-motors. An RC receiver is also located at the back. The role of the RC receiver is to remotely switch on/off the computer system and to activate the opening of parachute, which was manually done by a ground crew during testing.

 

Vision-based Guidance

Vision-based guidance approach was taken. In this concept, the onboard camera image is transmitted to the ground station. Based on the display of this image, an operator identifies and designates the target area. This approach has a few advantages. Depending on situations, as the vehicle gets close to the drop area the ground operator can change the specific drop zones in real time. For a situation where the exact GPS coordinates are not available, the vehicle can be dropped based on rough idea for the location of the drop zone, and as the vehicle gets close to the concerned area a specific location can be selected for target position in real time by the ground operator.

The pixel coordinates on the screen is then transmitted back to the drop vehicle. This information is combined with onboard inertial sensors in extended Kalman filter for navigation. The guidance strategy is to hold the target image somewhere on the camera image, for example, at the center. If the body can generate enough side force and lift and the target image is held on the camera image throughout the drop period, then precision drop is guaranteed.

Drop Testing from Aircraft

The aircraft used for the drop tests is shown on the right photo. It was called "Buckley Aircraft" to follow the name, Jack Buckley, who built this vehicle. His son, Mitch Buckley, was a primary pilot during our testings. It is a rather large (wing span ~ 5 m) RC piloted plane with an engine power of 15HP. It has a payload capacity of up to 50 pounds and it can fly for over 30 minutes.

Release mechanisms were developed for reliable release of the drop vehicle. Two of them were installed under the wings of the aircraft. A simple latch system was devised that is actuated by RC servos, which was remotely activated by a ground crew.

In the fall of 2003, the first main focus of testing was to check our release mechanism and parachute opening device. Here is a video from one of our drop test. The drop vehicle was release at the altitude of about 350 meters. It fell for about 10 seconds and reached its terminal speed of about 60 m/s. The parachute was deployed for safe landing at the end.

When the drop vehicle was completed with all the avionics and control logics, several attempts were made as a final demonstration. But we had to struggle. It turned out that after each drop, the mechanical and electrical components were often damaged from the impact of the parachute opening or sometime from hard landing on the ground. This damage frequently resulted in a rather nasty problem - reduction of communication ranges for both the image transmission and data exchange between the drop vehicle computer and ground station. Furthermore, very unfortunately, the aircraft crashed. In the middle of the flight, pilot lost the total control of the aircraft and the plane dived into a very dense wood area in Shirley, MA.

 

Drop Testing from Building

A few drop tests were attempted, instead, from the roof of Green Building at MIT. The height of the Green Building is about 87 meters. Although it is not an enough altitude, we tried.

 

Test Setup

In order to have enough clearance against the wall of the building, an extension structure was built before the test. The length of the extension bar was about 15 ft. And the release mechanism was attached at the end of the extension bar. The bar was designed to swing about a pivot point. So, when the drop vehicle was attached to the release device the bar was rotated about 90 degrees toward outside of the building for maximum clearance.

 

Roll Control

The video on the left shows the onboard camera view from one of our drop test. The white square box is our 4m x 4m testing target area.

The initial target pixel location at the upper portion of the camera image created a pitch-up control reaction right after the drop. But this motion was mixed with the initial roll motion from our imperfect release mechanism. The combined effect created a swing in the target pixel position. But after this initial transition period, the vehicle motion was stabilized by the roll control and the target was moved toward the center of the image space. But due to short altitude, there was not enough time for the vehicle to correct its position and it missed the target.

 

Body Orientation toward Target area

The two videos on the left show the outside view and the roof top view from one of our drop test. From these videos you can see that the vehicle pointed its body orientation toward the target area during the free-falling until the parachute opens. It drifted away when the parachute opens.


Summary & Recommendation

Although we were not very successful at the drop test using aircraft, we demonstrated the viability of our chosen concept from the drop test at the building, by showing that the vehicle motion was stabilized enough for a ground operator to trace the target and by verifying that the vehicle controlled its body orientation to point toward the target.

In the current vehicle design the lift/side-force relies on the cylindrical body shape. In order to have more control authority and larger down-range capability, lifting body design can be considered.

Other nice photos

Acknowledgement

I did this project with Josh Torgerson, Hyungil Ahn, and Thomas Jones. We had lots of fun together and I learned a lot from them. I also thank Prof. Deyst and Sean George for their advices and guidance throughout the project. Mitch Buckley was the pilot for our aircraft. Thanks to him, we were able to just concentrate on our testing without worrying about the aircraft operation.


Back to Home